Just like I did last year, I now want to go over the results of 4 Automated Forex strategies that have been discussed on this site in the past. The results last year were pretty good across the board, putting to rest the notion that "Forex robots don't work... wah-wah-wah".
Of course I have discussed several other Forex trading strategies in the past. Most of the time just exposing trash, scams and over curve-fitted junk, especially during the early life stages of this site. Just like I said last year, after a while, I realized the amount of garbage was insanely beyond my reach and it didn't make much sense to keep exposing charlatans and endlessly picking fights.
So here we are, focusing on robust Forex trading strategies with long profitable track records and great likelihood of being profitable in the future. I say great likelihood due to the fact that they exploit a very simple principle of the markets: trends exist. These are systems with simple rules, few degrees of freedom and strong and wide sets of profitable parameters, which suggest low degrees of curve-fitting.
I purchased high quality data (FXDD broker) to run all these tests on the MetaTrader4 platform. Bid-Ask spread used of 2 pips in the simulations, which is higher than what most brokers offer nowadays, putting our tests at a disadvantage. I adjusted the time zone of each quote by adding three hours so that the weeks only have 5 daily candles (Monday through Friday), all of them full 24 hour candles, instead of the shortened Sunday night and shortened Friday.
The currency I'm using, just as last year, is the Euro, or else this article would be and endless orgy of back-tests. Let's look at the results. (By clicking on the name of a strategy, you can visit the original article where it was first discussed)
EURUSD Donchian breakout strategy with time based exit
This strategy enters on a 70 day Donchian Channel breakout, sets a Stop Loss 2 ATRs away and a Profit Target 3.2 ATRs away. If neither Stop Loss nor Profit Target is achieved in 6 days, the system will simply automatically exit the position as it assumes the initial momentum after the break out may be fading. 3% capital allocation per trade on this test.
(Click on images to enlarge)
Solid +38% return in 2015. Very strong year. This is on the back of a +16.79% return in 2014. The Max draw-down in the equity curve was 23.66% (Like I have said in the past, Metatrader over estimates the draw-down as it assumes floating profits and losses instead of just closed positions). 34 trades in total, 23 winners, 11 losers. This is a pretty aggressive system, and not every single year is a winner, but when it wins, it usually wins big.
50 Day Breakout system with 50 SMA - 100 SMA crossover confirmation
A strategy kindly shared by portfolio manager Pavel Kařízek, who trades this mechanical system on several currencies and futures. For simplicity, again we just show the results on the EURUSD in 2015 or else this article would be infinite.
Because the strategy is applied on so many instruments at the same time, Pavel trades each one with very low risk settings. Let's see how it did for the Euro in 2015, max risk per trade of only 1%.
(Click on images to enlarge)
A +1.79% return for 2015 (Euro only) on the back of a +3.86% return in 2014. Very small 1.39% draw-down experienced. Although this is a small return, take into account the insignificant draw-down and also the fact that this strategy is traded in a portfolio with many other instruments, including not only currencies, but also futures, metals and grains.
The Turtles Trading System (Method 2)
The legendary Turtles Trading system doesn't need an introduction. This beauty has been killing it since introduced back in the 80's. For more details, click on the link above and enjoy.
By the way I recently finished programming a prototype for Method 1 of the Turtles, which was insanely more complex to program. The results of Method 1, when applied to currencies leave a lot to be desired in respect with Method 2. It is profitable as well, just not as attractive, and highly correlated (draw-downs of method 1 tend to coincide with those of method 2, and this is why the Turtles used "either or"). I don't think the Method 1 Robot is even worth publishing.
Back to Method 2 in 2015. Starting from December 15, 2014 instead of January 1, 2015. This is because on January 1, there were positions open since December 15 that had not been closed yet. So, in order for those results to be included in 2015 (which is when the positions were closed) this run has to start a little earlier than Jan 1, 2015.
So, how was 2015 for the Turtles Trading system (method 2) applied to the EURUSD currency pair?
Simply another excellent year.
(Click on images to enlarge)
Very strong +22.80% return in 2015 after a spectacular +35.60% the previous year
Draw-down (closed positions only): 5.62%.
10 winners and 6 losers in 2015 (vs 7 winners and 15 losers in 2014, yet the winners were larger in 2014).
When combining 2014 and 2015 in a single run, the strategy has returned +62% in just these two years, with a maximum draw-down of 11.62%. Unbelievable how a system with simple rules for trend following is still killing it 3 decades later.
The Turtles Trading System is not the Holy Grail. It is in fact extremely hard to trade and very few individuals would be able to mentally handle it. During years of good trends it will kill it, but in other years it can and will have extended draw-downs in both depth and duration.
LT Trend Sniper
My Darlin. From yours truly, the LT Trend Sniper system: my premium strategy designed for following 2 - 3 month long trends on currencies such as the Euro (EURUSD), the Sterling pound (GBPUSD) and the Australian dollar (AUDUSD). I have recently tested it with success in Gold as well, as I could finally get my hands on decent quality 1-minute Gold data going back to January 1, 2001.
So, here's the Sniper with the EURUSD in 2015. Also starting the test on December 15 to avoid disregarding positions that were closed in 2016 but which were initiated in 2015.
(Click on images to enlarge)
+8.20% return which is near its historical yearly average. The result differs a bit from what I achieved in my personal account because I was using a small $2,000 balance. Position sizes on my small account could not target the exact desired risk with the same granularity as a large account would have, like the $100,000 used in this test. Max Draw-down: 7.02%.
Many people have asked me whether I prefer the Sniper or the Turtles system and my answer is: The Sniper. My opinion is based on the Average Annual Return to Max-Draw-down ratio. The Sniper has historically obtained better returns for the draw-downs that it experiences. We could increase the position size, to match the historical returns of the Turtles and it would experience way smaller draw-downs at those risk levels.
When it comes to the portfolio of EURUSD + AUDUSD + GBPUSD, the system had a mediocre result, mainly due to the under-peformance of the Sterling Pound (GBPUSD) which had a -11.62% year and brought the portfolio down 1.02% for the year 2015.
Other symbols like Gold (XAUUSD) and the Canadian Dollar (USDCAD) had positive results of +1.50% and +14.70% respectively.
Euro +8.20%
Australian Dollar +2.19%
Sterling Pound -11.62%
Canadian Dollar +14.70%
Gold +1.50%
All those instruments combined in the same portfolio would have resulted in a +15.33% return with a 12.67% draw-down.
I have met so many traders entirely opposed to 100% automated trading systems. If automated trading systems truly don't have a chance, then why do those same people always complain that the markets are dominated by "algos".
Profitable automated trading is possible. It has been done. It is happening every day in the markets and it will continue to happen.
But, just like I said last year, be careful. There is an extraordinary amount of useless garbage out there. Especially in the world of retail Forex trading where automation is more prevalent and unscrupulous vendors distributing curve-fitting junk just to make a quick buck by selling the get rich overnight dream.
Let's let the "algos" work, and see how they perform in 2016. It would be cool to show extended track records of fully automated systems over 10 or 20 years. Let's see if I'm alive by then. As long as I am, results will continue to be shared year after year no matter what.
LT
Of course I have discussed several other Forex trading strategies in the past. Most of the time just exposing trash, scams and over curve-fitted junk, especially during the early life stages of this site. Just like I said last year, after a while, I realized the amount of garbage was insanely beyond my reach and it didn't make much sense to keep exposing charlatans and endlessly picking fights.
So here we are, focusing on robust Forex trading strategies with long profitable track records and great likelihood of being profitable in the future. I say great likelihood due to the fact that they exploit a very simple principle of the markets: trends exist. These are systems with simple rules, few degrees of freedom and strong and wide sets of profitable parameters, which suggest low degrees of curve-fitting.
I purchased high quality data (FXDD broker) to run all these tests on the MetaTrader4 platform. Bid-Ask spread used of 2 pips in the simulations, which is higher than what most brokers offer nowadays, putting our tests at a disadvantage. I adjusted the time zone of each quote by adding three hours so that the weeks only have 5 daily candles (Monday through Friday), all of them full 24 hour candles, instead of the shortened Sunday night and shortened Friday.
The currency I'm using, just as last year, is the Euro, or else this article would be and endless orgy of back-tests. Let's look at the results. (By clicking on the name of a strategy, you can visit the original article where it was first discussed)
EURUSD Donchian breakout strategy with time based exit
This strategy enters on a 70 day Donchian Channel breakout, sets a Stop Loss 2 ATRs away and a Profit Target 3.2 ATRs away. If neither Stop Loss nor Profit Target is achieved in 6 days, the system will simply automatically exit the position as it assumes the initial momentum after the break out may be fading. 3% capital allocation per trade on this test.
(Click on images to enlarge)
Solid +38% return in 2015. Very strong year. This is on the back of a +16.79% return in 2014. The Max draw-down in the equity curve was 23.66% (Like I have said in the past, Metatrader over estimates the draw-down as it assumes floating profits and losses instead of just closed positions). 34 trades in total, 23 winners, 11 losers. This is a pretty aggressive system, and not every single year is a winner, but when it wins, it usually wins big.
50 Day Breakout system with 50 SMA - 100 SMA crossover confirmation
A strategy kindly shared by portfolio manager Pavel Kařízek, who trades this mechanical system on several currencies and futures. For simplicity, again we just show the results on the EURUSD in 2015 or else this article would be infinite.
Because the strategy is applied on so many instruments at the same time, Pavel trades each one with very low risk settings. Let's see how it did for the Euro in 2015, max risk per trade of only 1%.
(Click on images to enlarge)
A +1.79% return for 2015 (Euro only) on the back of a +3.86% return in 2014. Very small 1.39% draw-down experienced. Although this is a small return, take into account the insignificant draw-down and also the fact that this strategy is traded in a portfolio with many other instruments, including not only currencies, but also futures, metals and grains.
The Turtles Trading System (Method 2)
The legendary Turtles Trading system doesn't need an introduction. This beauty has been killing it since introduced back in the 80's. For more details, click on the link above and enjoy.
By the way I recently finished programming a prototype for Method 1 of the Turtles, which was insanely more complex to program. The results of Method 1, when applied to currencies leave a lot to be desired in respect with Method 2. It is profitable as well, just not as attractive, and highly correlated (draw-downs of method 1 tend to coincide with those of method 2, and this is why the Turtles used "either or"). I don't think the Method 1 Robot is even worth publishing.
Back to Method 2 in 2015. Starting from December 15, 2014 instead of January 1, 2015. This is because on January 1, there were positions open since December 15 that had not been closed yet. So, in order for those results to be included in 2015 (which is when the positions were closed) this run has to start a little earlier than Jan 1, 2015.
So, how was 2015 for the Turtles Trading system (method 2) applied to the EURUSD currency pair?
Simply another excellent year.
(Click on images to enlarge)
Very strong +22.80% return in 2015 after a spectacular +35.60% the previous year
Draw-down (closed positions only): 5.62%.
10 winners and 6 losers in 2015 (vs 7 winners and 15 losers in 2014, yet the winners were larger in 2014).
When combining 2014 and 2015 in a single run, the strategy has returned +62% in just these two years, with a maximum draw-down of 11.62%. Unbelievable how a system with simple rules for trend following is still killing it 3 decades later.
The Turtles Trading System is not the Holy Grail. It is in fact extremely hard to trade and very few individuals would be able to mentally handle it. During years of good trends it will kill it, but in other years it can and will have extended draw-downs in both depth and duration.
LT Trend Sniper
My Darlin. From yours truly, the LT Trend Sniper system: my premium strategy designed for following 2 - 3 month long trends on currencies such as the Euro (EURUSD), the Sterling pound (GBPUSD) and the Australian dollar (AUDUSD). I have recently tested it with success in Gold as well, as I could finally get my hands on decent quality 1-minute Gold data going back to January 1, 2001.
So, here's the Sniper with the EURUSD in 2015. Also starting the test on December 15 to avoid disregarding positions that were closed in 2016 but which were initiated in 2015.
(Click on images to enlarge)
+8.20% return which is near its historical yearly average. The result differs a bit from what I achieved in my personal account because I was using a small $2,000 balance. Position sizes on my small account could not target the exact desired risk with the same granularity as a large account would have, like the $100,000 used in this test. Max Draw-down: 7.02%.
Many people have asked me whether I prefer the Sniper or the Turtles system and my answer is: The Sniper. My opinion is based on the Average Annual Return to Max-Draw-down ratio. The Sniper has historically obtained better returns for the draw-downs that it experiences. We could increase the position size, to match the historical returns of the Turtles and it would experience way smaller draw-downs at those risk levels.
When it comes to the portfolio of EURUSD + AUDUSD + GBPUSD, the system had a mediocre result, mainly due to the under-peformance of the Sterling Pound (GBPUSD) which had a -11.62% year and brought the portfolio down 1.02% for the year 2015.
Other symbols like Gold (XAUUSD) and the Canadian Dollar (USDCAD) had positive results of +1.50% and +14.70% respectively.
Euro +8.20%
Australian Dollar +2.19%
Sterling Pound -11.62%
Canadian Dollar +14.70%
Gold +1.50%
All those instruments combined in the same portfolio would have resulted in a +15.33% return with a 12.67% draw-down.
I have met so many traders entirely opposed to 100% automated trading systems. If automated trading systems truly don't have a chance, then why do those same people always complain that the markets are dominated by "algos".
Profitable automated trading is possible. It has been done. It is happening every day in the markets and it will continue to happen.
But, just like I said last year, be careful. There is an extraordinary amount of useless garbage out there. Especially in the world of retail Forex trading where automation is more prevalent and unscrupulous vendors distributing curve-fitting junk just to make a quick buck by selling the get rich overnight dream.
Let's let the "algos" work, and see how they perform in 2016. It would be cool to show extended track records of fully automated systems over 10 or 20 years. Let's see if I'm alive by then. As long as I am, results will continue to be shared year after year no matter what.
LT
Go to the bottom of this page in order to see the Legal Stuff
No comments:
Post a Comment